Automatic driving implementation method and road watching system

Abstract

The present invention provides a method for realizing automatic driving, comprising: acquiring perception information, wherein the perception information includes at least one of environmental information, map information, and traffic control information; obtaining automatic driving related data based on a fusion processing of the perception information with at least one of a road management status and drum beat data; based on the automatic driving related data, performing at least one of road condition prediction, path planning, and driving decision-making, executing driving control, and realizing automatic driving.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to an autonomous driving implementation method and a road watching system. Background Art

[0002] With the rapid development of science and technology, unmanned autonomous driving has entered people's horizons. As research gradually deepens, autonomous driving will play a vital role in people's work and life.

[0003] The technology of self-driving cars has become increasingly mature, but there are still many technical challenges to be overcome before they can be put on the road. One of the challenges is that the central control computer of a self-driving car requires a huge amount of data input to determine the car's driving route, turning, braking, acceleration, avoiding pedestrians, complying with traffic rules and other actions.

[0004] Therefore, there is a need to provide a safer method for implementing autonomous driving and a road watching system. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for realizing automatic driving, in order to solve the technical problems existing in the background technology.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A method for implementing automatic driving, comprising:

[0008] Acquiring perception information, where the perception information includes at least one of environmental information, map information, and traffic control information;

[0009] Obtaining autonomous driving related data based on a fusion process of the perception information with at least one of a road management state and drum beat data;

[0010] Based on the autonomous driving related data, predictions, path planning, driving decisions, etc. are performed to determine the execution of driving control and realize autonomous driving.

[0011] In some embodiments, the obtaining of autonomous driving related data based on the fusion processing of the perception information with at least one of the management status of the road and the drum data includes:

[0012] An image-like grid data map obtained by gridding the road;

[0013] Based on the grid data graph, an image algorithm is used to obtain a grid area related to the road surface as a ROI area; wherein each grid in the ROI area corresponds to the right to use one grid unit of the road surface;

[0014] Based on a preset calibration algorithm, the usage status of each grid is calibrated in combination with the perception information and the management status of the road; the road edge and unavailable grid cells are filled with busy state or non-state machine specific values; available grid cells are filled with Idle state specific values ​​to form a static grid map;

[0015] Based on the real-time road conditions and traffic participant information in the service area, a preset algorithm is used to distinguish grids that meet the first preset condition, and the grids are calibrated using the same specific values ​​as the static grid map; wherein the road surface grids corresponding to the traffic participants are calibrated using the busy state specific values; the grids corresponding to the road conditions that meet the first preset condition are calibrated using the busy state specific values ​​or non-state machine specific values ​​to form a dynamic grid map; wherein the first preset condition may be a situation that is not suitable for traffic, such as a curb or isolation belt, as well as a situation that affects road traffic, such as collapse, deep wading, ice, dense fog, dense smoke, fire, strong crosswinds, etc.;

[0016] Based on the perception information and the management status of the road, a traffic control grid map is generated, and traffic control signs are refined for each grid. The grid map of the service area is formed after being merged with the static grid map and the dynamic network map and transmitted to the on-board unit OBU of the EV;

[0017] According to a preset drumbeat or a preset cycle, adjust the usage status or time slice timing data of each grid in the grid map of the maintenance service area;

[0018] Based on the movement of EVs, the occupancy and release of the unit grid road surface in the grid map of the service area are updated, and the updated grid map of the service area is fed back to the OBU of the EV to form an information closed loop.

[0019] The OBU of the EV generates or updates a driving plan based on the updated grid map of the service area.

[0020] In some embodiments, the calibrating the usage status of each grid includes:

[0021] The available road surface grid is initialized to the Idle state. When a traffic participant enters the grid road surface, the corresponding grid is used and the state is switched to the busy state. Other traffic participants learn that the grid is occupied and no longer use the grid. After the OBU planning decision, they no longer enter the road surface corresponding to the grid to avoid accidents.

[0022] When a traffic participant walks out of the grid road surface, the state switches to blocked, the grid is locked and waiting to be released, other participants cannot enter, and the timing starts at the same time; drum count or system seconds, 100 milliseconds and other cycles can be used for timing. In order to ensure safety and maintain a safe distance, other participants will know that the grid is blocked and will not use the grid. After the OBU planning decision, they will no longer enter the road corresponding to the grid, reducing the probability of accidents;

[0023] After a certain period of time, the traffic participants have gone far away, the road surface corresponding to the grid is safe, and it returns to the Idle state and can be used by other traffic participants.

[0024] In some embodiments, the calibrating the usage status of each grid includes:

[0025] The available road surface is calibrated using a state machine, and the unavailable road surface is calibrated using a busy state or a value different from the state machine.

[0026] In some embodiments, in the static grid map and the dynamic grid map, the data of each grid is 8-bit data; the specific value of the state machine used contains three states and an unavailable specific value, which is represented by 2 bits.

[0027] At the same time, the present invention also discloses a road-watching system for realizing automatic driving, comprising:

[0028] A perception unit, used to provide information detection of at least one of the environment, road surface, and traffic participants, and obtain perception data;

[0029] A computing unit, used to implement at least one computing service selected from the group consisting of perception data fusion, map data fusion, communication data management, security data management, and edge computing;

[0030] Communication unit, used to exchange information with ITS, EV equipped with communication system, and other equipment;

[0031] The positioning unit is used to locate the main body and provide positioning services for devices such as EVs involved in autonomous driving;

[0032] A clock unit, used to provide at least one of the time series related services such as clock synchronization, high-precision time stamp, and drum beats;

[0033] A storage unit, used to store at least one of the non-volatile data such as engineering drawings, maps, operation logs, configuration parameters, etc., and to provide the computing unit with at least one of the volatile data such as operation memory, data storage, and intermediate result storage services;

[0034] A security unit, used to provide at least one security service among encryption and decryption, communication authentication, backup voting and fault tolerance services, and equipment security monitoring services;

[0035] The management unit is used to provide at least one management service among configuration management, maintenance management operation statistics, status monitoring, fault warning and upgrade management services.

[0036] In some embodiments, it also includes:

[0037] System interface, used to provide an interface for connecting to external systems.

[0038] In some embodiments, it also includes: a power supply unit for providing power management and working power.

[0039] In some embodiments, the system adopts a chain-like overlapping arrangement to back up each other, thereby improving the reliability and robustness of the system. After the road-tracking system is cascaded, the overlapping areas can be mutually verified, and distributed computing can also be performed.

[0040] In some embodiments, when the positioning unit fails, a preset algorithm is used to achieve positioning based on at least two eNBs of the system, and the preset algorithm can be trangle + similar algorithm for precise positioning. For example, after the dedicated positioning unit of the road sign and smart EV fails, trangle + similar algorithm can be used for precise positioning based on more than two road signs.

[0041] At the same time, the present invention also discloses a device for implementing the method for implementing automatic driving, the device comprising at least one processor and at least one memory;

[0042] The at least one memory is used to store computer instructions;

[0043] The at least one processor is configured to execute at least part of the computer instructions to implement any of the aforementioned methods.

[0044] At the same time, the present invention also discloses a computer-readable storage medium, which stores computer instructions. When the computer instructions are executed by a processor, any of the above-mentioned methods is implemented.

[0045] Beneficial Effects

[0046] Compared with the prior art, the present invention has the following significant advantages:

[0047] The automatic driving realization method and the road-watching system of the present invention integrate the operating system scheduling algorithm, manage the road surface, reduce the probability of traffic accidents, and provide a new solution for automatic driving. At the same time, the operating system RM grid map is compatible with the automatic driving planning and decision map. After simple processing, it can be used by the vehicle's OBU, reducing the complexity of the vehicle algorithm and improving the system's operating efficiency. By introducing synchronous drum beats (tick), attempts are made to solve the problem of inconsistent control steps of each EV, improve traffic efficiency, and reduce the probability of traffic accidents. Road watching is used to share holographic fresh environmental data for each vehicle, reduce system complexity, improve reliability, and improve system real-time performance; it provides the possibility of saving detection equipment for intelligent vehicles and reducing vehicle energy consumption and production costs. And the road watching service area overlaps and backs up to improve system reliability. Engineering drawings can be used to reduce dependence on high-precision maps and reduce the dependence of unmanned vehicles on national geographical features. And the scheduling accurate to each grid unit is conducive to the refined management of the road surface and improves the compatibility of the automatic driving system. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is a schematic diagram of a static grid map of a road involved in this embodiment;

[0049] Figure 2 is a schematic diagram of a dynamic grid map of a road involved in this embodiment;

[0050] Figure 3 A schematic diagram of a road RM graph with time slice counts according to an embodiment of the present invention;

[0051] Figure 4 A schematic diagram of a static grid diagram of an intersection involved in an embodiment of the present invention;

[0052] Figure 5 A schematic diagram of a dynamic grid diagram of an intersection involved in an embodiment of the present invention;

[0053] Figure 6 A schematic diagram of an intersection RM diagram including time slice counts according to an embodiment of the present invention;

[0054] Figure 7 A schematic diagram of a scenario of a pending line indication problem to be solved in an embodiment of the present invention;

[0055] Figure 8 A schematic diagram of the arrangement of a road monitoring system according to an embodiment of the present invention;

[0056] Fig. 9 A schematic diagram of a road-watching system according to an embodiment of the present invention;

[0057] Fig.10 A schematic diagram of a closed-loop information flow involved in an embodiment of the present invention;

[0058] Fig.11 A schematic diagram of a state machine for reference when calibrating a grid according to an embodiment of the present invention;

[0059] Fig.12 It is a schematic diagram of positioning of a base station and an EV involved in an embodiment of the present invention. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0061] On the contrary, the present application covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present application as defined by the claims. Further, in order to make the public have a better understanding of the present application, some specific details are described in detail in the detailed description of the present application below. Those skilled in the art can fully understand the present application without the description of these details.

[0062] The following will be combined Figure 1-12 A method for realizing an automatic driving system according to an embodiment of the present application is described in detail. It should be noted that the following embodiments are only used to explain the present application and do not constitute a limitation on the present application.

[0063] For the convenience of description, the following abbreviations are used:

[0064] C-ITS: cooperative intelligent transportation systems

[0065] LTE: Long Term Evolution

[0066] VRU; Vulnerable Road User

[0067] EV: Equipped Vehicle

[0068] NV: Normal Vehicle without communication system

[0069] Evolved Node B (E-UTRAN Node B)

[0070] V2X: Vehicle to everything

[0071] V2P: Vehicle to Pedestrian

[0072] V2V: Vehicle to Vehicle

[0073] V2I: Vehicle to Infrastructure

[0074] V2N: Vehicle to Networks

[0075] UE:User Equipment

[0076] LT-V2X: LTE-based V2X vehicle wireless communication technology (LTE Vehicle to Everything)

[0077] C-V2X: V2X wireless communication technology based on cellular communication (cellular vehicle-to-everything)

[0078] NR-V2X: V2X wireless communication technology based on new network communication technology (New radio vehicle-to-everything)

[0079] DSRC: Dedicated Shortrange Communication

[0080] OBU: On-Board Unit

[0081] RSU: Road Side Unit

[0082] RMU: Road Monitor Unit

[0083] SPAT: Signal Phase and Timing Message

[0084] SSM: Sensor Sharing Message

[0085] TPM: Traffic Probe Message

[0086] VIR: Vehicle Intention And Request UWB: UltraWide Band

[0087] GNNS: Global Navigation Satellite System

[0088] ITS: Intelligent Transportation System

[0089] ROI: region of interest

[0090] FCFS: First Come First Served

[0091] SLAM: Simultaneous Localization and Mapping

[0092] MEC: Multi-access / Mobile Edge Computing

[0093] TOA: time of arrival

[0094] TDOA: Time Difference 0f Arrival

[0095] OTA: Over-the-Air Technology

[0096] The purpose of the present invention is to provide an automatic driving method, which utilizes the scheduling of an operating system to allocate road use rights, reduces the complexity of the system, and accelerates the realization of an intelligent connected transportation system.

[0097] The basic principles of the method are as follows:

[0098] Principle 1. Whether it is driving or autonomous driving, its essence is the allocation and use of the right of way of a single road. When to obtain, use, and return the right of way is the core of driving and autonomous driving. The right of way is unique. Only one traffic participant can use it at the same time. This is very similar to the unique resources in the computing system. If the road is regarded as a hardware resource in the computing system, and the traffic participants are regarded as a thread that shares the use of the hardware resource, the right of way is only a period of use of the road. This is very similar to the operating system, but it is a special case that cannot be interrupted or deprived during the use period.

[0099] Principle 2: In ancient times, in order to maintain the formation, soldiers used drum beats to coordinate their movements. Soldiers moved in unison to avoid interference between soldiers. In autonomous driving, since the participants do not have similar coordination, they often move in unison, resulting in unnecessary braking, lane changes, overtaking and other actions, which reduces the efficiency of road traffic. By using a unified drum beat, each participant moves to the rhythm of the drum beat, and a perfect square formation similar to a military parade can be created, improving the efficiency of road traffic.

[0100] The solution to the above-mentioned problem is to integrate the environmental information, maps, traffic control information, etc. collected by the perception layer, and superimpose the road management status and drum data to form unified data for direct use by OBU path planning, prediction, decision-making, control and other algorithms.

[0101] The specific plan is as follows:

[0102] On the basis of the traditional LTE-type RSU, a perception layer is added to perceive the road environment and the dynamic status of each participant; ESU is added to provide edge computing (MSC) to complete the fusion of information such as environmental information, maps, and traffic control information; and road management status, drum data generation, superposition scheduling and maintenance services are completed. The new device is essentially different from the RSU. This device with perception function that provides edge computing, positioning, monitoring and control services for the autonomous driving system is redefined as a Road Monitor Unit, hereinafter referred to as RMU.

[0103] RMU is a road infrastructure and its relationship in the autonomous driving system is similar to that of RSU / SPAT. The relationship diagram and reference hardware architecture diagram are as follows: Fig. 9 As shown in the figure, RMU consists of a computing unit, a sensing unit, a communication unit, a positioning unit, a clock unit, a security unit, a storage unit and a management unit.

[0104] The computing unit includes but is not limited to secure computers, GPUs, FPGAs, etc. for edge computing (SC). It is the central hub of the RMU and realizes functions such as perception data fusion, map data fusion, communication, security, and management.

[0105] The perception unit includes but is not limited to cameras, LiDAR, millimeter-wave radar, infrared sensors, temperature and humidity, rain, snow, fog, wind, water level, collapse, cliff, ice, earthquake and other perception sensors. It provides RMU with detection and testing of traffic elements such as environment, road surface, and traffic participants, and provides data sources for RMU computing nodes.

[0106] The communication unit includes but is not limited to C-V2X, NR-V2x, PC5, DSRC and other communication devices for exchanging information with EVs equipped with communication systems.

[0107] The positioning unit includes but is not limited to GNSS and UWB equipment, which is used to provide high-precision positioning services for EVs within the RMU service range.

[0108] The clock unit provides clock synchronization, high-precision time stamping, drum beat and other services for the RMU.

[0109] The storage unit is used to store engineering drawings / maps, operation logs, configuration parameters and other data for the RMU, and to provide the computing unit with operation memory, data storage and intermediate data storage services.

[0110] The security unit provides the RMU with encryption and decryption, communication authentication, two-out-of-two, three-out-of-two and other similar backup voting and fault tolerance, equipment security monitoring and other services.

[0111] The management unit provides services such as configuration management, maintenance management, operation statistics, status monitoring, fault warning and upgrade management.

[0112] The system interface provides interfaces with the autonomous driving platform, interfaces between RMUs, interfaces for auxiliary equipment, etc. The interface of the autonomous driving platform is used to communicate with the autonomous driving platform, edge cloud, traffic management platform, etc. The RMU interface is used to interact with other RMU interfaces or form a distributed computing network; the auxiliary interface is used to communicate with traditional traffic signal equipment, expand perception and other equipment.

[0113] The power supply unit is used for power management of the RMU and provides working power for each unit of the RMU.

[0114] Compared with the smart car solution, the vehicle-road collaboration using RMU is greatly reduced; but its cost is still higher than that of general RSU. In order to make full use of the bird's-eye view advantage of the installation structure such as the pole, make full use of the detection resources of RMU cameras, lidar, microwave radar, etc., expand the V2X service range, and improve the safety of RMU. It is recommended to install RMU in the central isolation belt and safety guardrail of the road. The central layout also facilitates the construction of RMU engineering, RMU communication and the realization of self-healing loop, and is beneficial to the safety protection of RMU equipment.

[0115] At the same time, it is recommended that RMUs be arranged in a chain-like overlapping manner. The service areas of RMUs overlap with each other, which plays a role of redundant backup and improves the reliability and robustness of the system. The redundant data sources obtained by EVs in the service area facilitate OBUs to adopt voting mechanisms such as two-out-of-two and three-out-of-two, thereby improving the fault tolerance of the system. Figure 8 As shown. As we all know, the effective communication coverage of radio is a circle with the antenna as the center. At the same time, the effective perception range of laser radar, millimeter wave radar, surround view camera, etc. is also a circle with the installation position as the center. For the convenience of description, we assume that the effective communication coverage range and the effective perception range are the same, collectively referred to as the effective coverage range. In this figure, A, B, and C are arranged in a chain-like overlapping manner, and the effective coverage range of A overlaps with the effective coverage range of B, and their intersection area is hereinafter referred to as the first area, and the effective coverage range of B overlaps with the effective coverage range of C, and their intersection area is hereinafter referred to as the second area; under normal circumstances, the first area is served by A and B, and the second area is served by B and C. When the B node fails, the first area it serves is continued to be served by the A node, and the second area is continued to be provided by the C node, realizing the backup of the node. The fault backup of other nodes is similar to it, and will not be repeated. On the other hand, the EV in each area receives data from two nodes, which is mutually verified with the data perceived by the EV's own OBU, reducing the probability of misjudgment at each perception layer and improving reliability. The train safety computer system can be referred to, and similar voting structures such as two-out-of-two and three-out-of-two can be adopted to improve the fault tolerance of the system.

[0116] The following significant effects can be achieved by using the overlapping chain arrangement of RMU in the middle of the road:

[0117] RMU is installed in a fixed manner, and its position can be accurately located, providing a positioning reference point for EV, which is convenient for improving positioning accuracy. Its position is fixed and can be obtained by GNNS, or by high-precision measurement during construction, saving GNSS equipment and related resources.

[0118] RMU is fixed, and compared with EV's OBU, it is more convenient to obtain energy, has higher supply efficiency, has greater room for performance improvement, and is more convenient for system deployment and maintenance such as OTA. RMU serves as a perception layer shared by many unmanned vehicles, which helps unmanned vehicles to cut sensor components, reduce manufacturing difficulty, save system costs, and accelerate the realization and deployment of unmanned systems.

[0119] The RMU is fixed, which makes it easy to expand the detection of special conditions such as road wading, ice, fire, collapse, landslide, etc., and improve the safety and compatibility of the system. However, it is very difficult to expand on EV, and even requires redesign and development.

[0120] Usually, smart EVs need to conduct a lot of learning for different features and scenarios, and their algorithms are complex and inefficient. In contrast, RMUs are fixed and only focus on the detection, calculation, and decision-making of the environment, participants, and other elements within their service range, which facilitates the deployment of deep learning and improves its efficiency.

[0121] RMU is designed specifically for transportation. The time when EVs pass through its service area can be predicted. In theory, when no EVs pass through its service area, it can be put into sleep mode appropriately, thus achieving energy-saving effects and alleviating problems such as excessive energy consumption of 5G base stations.

[0122] The RMUs are arranged in the middle of the road, with EVs passing on both sides, which helps protect the RMUs and avoid certain sabotage activities. At the same time, each RMU is within the detection range of other RMUs, which helps protect the safety of the RMUs.

[0123] Central isolation area, where underground pipe network can be designed to provide power supply for RMU, communication fiber healing ring, etc. It is helpful to provide supporting facilities and avoid damage to the system by other construction. The autonomous driving system can also be physically isolated from the civil network to become a dedicated network to ensure the reliability of the autonomous driving system. It is convenient to build an RMU distributed computing network and improve the computing power of the system.

[0124] After the RMU service range overlaps, when high-precision positioning units such as UWB fail, more than two RMUs can be used to replace high-precision positioning. This plays the role of high-precision positioning hot backup and improves the robustness of the autonomous driving system. With the help of GNNS-IMU or electronic compass, the positioning algorithm based on RMU can use a simple triangle solution algorithm in analytic geometry.

[0125] The trangle+ algorithm can be used for, but not limited to, high-precision positioning of UWB and eNB base stations. The algorithm is as follows:

[0126] The distance between RMUs is short, and they are arranged in a chain-like overlapping pattern in the middle of the road, so the height difference between RMUs can be ignored. Using two adjacent RMUs as positioning base stations can achieve high-precision positioning. Fig.12 As shown, the coordinates of the two RMUs are like the two points A and B in the figure, which have become conditions. With the coordinates of the two points, the length c of the side AB can be easily obtained:

[0127]

[0128] With RMU A and RMU B as positioning base stations, the EV at point C obtains the distances b and a between CA and CB through TOA, TDOA and other technologies. Given the lengths of three edges and the coordinates of two vertices, the coordinates of the third vertex can be solved using the forward and reverse calculation formulas of the coordinates. The reference steps are as follows:

[0129] First, calculate the angle θ between point A and side BC. The formula is:

[0130]

[0131] Then, according to the coordinate inversion formula, calculate the azimuth angle α of the AB side relative to the Y axis AB :

[0132]

[0133] Among them, the result of the above formula is a range of (-90°, 90°), so it has certain limitations;

[0134] To this end, the universal formula of azimuth can be used to accurately calculate the azimuth within 360 degrees:

[0135] Az=pi*(1-sgn(dY) / 2)-Atn(dX / dY)

[0136] This formula does not need to determine the quadrant, just add 360 when the value is less than 0. Among them, sgn() is a sign function, if dX<0, its value is -1, dX>0, and dX=0, it is 0.

[0137] Next, calculate the azimuth angle α of the AC side AC

[0138] α AC =α AB ±θ

[0139] At this time, there are two solutions in the formula, and the two are symmetrical with respect to the AB side. In the real world, they are reflected as the two sides of the RMU. At this time, the following method can be used to determine:

[0140] The coordinates obtained by the GNSS+IMU system on the EV are determined, the azimuth of the EV electronic compass is matched with the lane, and the EV perception image is matched and determined.

[0141] In the figure, the azimuth of the AC side needs to be subtracted from θ: α AC =α AB -θ

[0142] Finally, according to the coordinate calculation formula, the coordinates of point C are calculated as follows:

[0143]

[0144] RMU uses engineering drawings, maps, high-precision maps and perception units to verify data, and grids the RMU service area according to certain units. Each grid is like a pixel in an image. After gridding, the RMU service area is like a digital image. The number of grids increases with the increase of grid resolution. For the convenience of description, it is agreed that the specification uses a lane width for gridding. Each grid data is represented by 8 bits, and the central isolation belt, each traffic participant, curb, etc. are all represented by a grid. By using general image normalization, edge detection and other related algorithms, the ROI area related to the road surface can be obtained.

[0145] In the ROI area, specific values ​​are used for filling. The road edges and unavailable grid cells are filled with (busy) state or non-state machine specific values; available grid cells are filled with Idle state specific values ​​to form a static grid map. The static grid map of the road is as follows: Figure 1 The static grid diagram of the intersection is shown in Figure 4 As shown, the specific values ​​in the figure adopt the time slice scheduling specific values, and this specific value is adopted in all figures.

[0146] Each grid in the ROI area corresponds to the right to use one grid unit of the road surface. In the operating system, state machines or time slices are often used to mark the right to use hardware. The right to use the gridded road is also calibrated in a similar way. The reference state machine is as follows: Fig.11 As shown. The available road surface grid is initialized to the Idle state. When the traffic participant enters the grid road surface, the corresponding grid is used, and the state machine switches to the busy state. Other participants learn that the grid is occupied and no longer use the grid. After the OBU planning decision, they can no longer enter the road surface corresponding to the grid to avoid accidents. When the traffic participant walks out of the grid road surface, the state machine switches to blocked, the grid is locked, waiting to be released, and the system starts the timer to start timing. The drum count or the system's seconds, 100 milliseconds and other cycles can be used for timing; in order to ensure safety and maintain a safe distance, other participants learn that the grid is blocked and do not use the grid. After the OBU planning decision, they will no longer enter the road surface corresponding to the grid to reduce the probability of accidents; after a certain period of time, the user has walked far away, the road surface corresponding to the grid is safe, and it returns to the Idle state and can be used by other participants.

[0147] The available road surface is calibrated using a state machine, and the unavailable road surface is calibrated using a busy state or a value different from the state machine.

[0148] For time slice scheduling, the minimum value corresponds to the idle state, the maximum value corresponds to the busy state, and other values ​​correspond to the blocked state. Taking 8bit as an example, the grid corresponding to the available road surface can be initialized to 0x00 (idle state); when the traffic participant enters the grid road surface, the corresponding grid data is set to 0xFF (busy state), and the grid is occupied; when the traffic participant exits the grid road surface, the grid enters the blocked state, and the grid data begins to decrease at a certain period. When the value decreases to the minimum value, the grid road surface is safe and returns to 0x00 (idle state).

[0149] RMU's perception units, such as cameras, are usually installed on high poles, towers, or the tops of bridges and tunnels. With the advantage of high position, its effective detection range and effect are much better than vehicle-mounted perception, especially for the detection of "ghosts". Its fixed bird's-eye view is also conducive to calculation and fusion. Its algorithm complexity is lower than that of vehicle-mounted perception, and its efficiency is also significantly higher than that of vehicle-mounted perception.

[0150] RMU obtains the real-time road conditions and traffic participant information in the service area; through relevant algorithms, it distinguishes traffic participants and grids involved in collapse, deep wading, ice, and dense fog; and uses the same method as static grids to Figure 1 The road surface grids corresponding to the participants are calibrated with the (busy) state specific values; the grids corresponding to road conditions such as collapse, deep wading, ice, dense fog, dense smoke, fire, and strong crosswinds are calibrated with the (busy) state specific values ​​or non-state machine specific values ​​to form a dynamic grid diagram.

[0151] It is worth mentioning that when the perception network detects prohibited conditions such as collapse, deep water, ice, thick fog, thick smoke, fire, strong crosswinds, etc., for safety reasons, it is recommended that all grids in the RMU service area be marked as occupied (busy), the drumbeats stop, and passing is prohibited; when the perception network detects conditions where speed can be reduced, such as shallow water, light fog, and light crosswinds, it is recommended that the RMU slow down the drumbeats and coordinate EVs to reduce speed. In addition, the RMU's perception unit can distinguish between VRUs, EVs, UEs, and animals that have strayed, and V2P and V2V are no longer necessary options. The dynamic grid diagram of the road is shown below. Figure 2 The dynamic grid diagram of the intersection is shown in Figure 5 As shown,

[0152] RMU obtains traffic control information detected by the perception unit, traffic control information from engineering drawings / maps, and traffic control information from the RMU coverage area managed and configured by the autonomous driving platform through algorithms. RMU can use the RSU-compatible V2I interaction protocol to pass it to the OBU of the EV, and the OBU will perform traffic control analysis; it can also generate a traffic control grid map, refine the traffic control signs for each grid, and pass it to the OBU of the EV after merging it with the static grid map and dynamic network map; the fused RM grid map is rich in information, and the OBU's path planning, decision-making, control and other algorithms only need to focus on the image algorithm. The direct transmission method is very simple and will not be described here. The following is an explanation of the reference fusion method:

[0153] In static grid maps and dynamic grid maps, the data of each grid is 8 bits of data. The specific value of the state machine used only needs to contain three states and unavailable specific values, and can be represented by 2 bits. When using time slices, although the counts used, it is often not necessary to count for a large amount of time to determine that the grid is safe and can be reused. Affected by wireless transmission, bit splicing and compression techniques can be used to reduce communication data and improve communication efficiency; with the application of new technologies such as NR-V2X, the communication rate has been greatly improved, and the use of full data has become a good option. The state machine and time slice specific values ​​can be compressed to 4 bits, and the remaining 4 bits are expanded for traffic information. For example, the state machine reference expansion is as follows:

[0154] Bit6-7 state machine specific value, Bit4-5 state machine specific value backup, Bit3 prohibit forward movement, Bit2 prohibit U-turn, Bit1 prohibit left movement, Bit0 prohibit right movement.

[0155] The time slice specific value reference extension is as follows:

[0156] Bit4-7 time counting, Bit3 prohibits going forward, Bit2 prohibits U-turn, Bit1 prohibits going left, and Bit0 prohibits going right.

[0157] RMU integrates static grid map and dynamic grid map, and integrates traffic control grid map into the full data to form the grid map of RM service area. According to the drum beat or a fixed period, it adjusts and maintains the usage status or time slice timing data of each grid. This road RM map with time slice count is as follows: Figure 3 As shown, the intersection RM diagram is as follows Figure 6 shown.

[0158] Added traffic light information grid reference diagram Figure 6As shown: the left and right sides of the intersection are allowed to pass, and the upper and lower roads are prohibited from passing when the red light is on. The grid cells corresponding to the pedestrian crossings on the left and right roads can be used for EV passage. The pedestrian crossings on the upper and lower roads are divided into pedestrian crossing occupied areas and shared areas. The pedestrian crossing occupied area is set to Busy state and EVs are prohibited from passing. The shared area of ​​the pedestrian crossing is used by EVs and pedestrians, and can be passed according to the participant FCFS mechanism.

[0159] The adoption of grids has refined road traffic management, while also improving the compatibility and scalability of traffic management. For example, in recent years, "waiting areas" have appeared on urban roads in my country. There are no standard traffic management signs for the use of these "waiting areas", and most of them use traffic control signs for reminders. Moreover, there has been no mature research and publication on these extended signs in autonomous driving. There are no relevant specifications in national and international autonomous driving standards, and in the standards that have been issued and are being formulated. Regarding "waiting areas", most of the announcements are made using information billboards. In single-vehicle intelligence, we can only rely on cameras to obtain announcement images and perform OCR on the images, and there are inconsistencies in the content of billboards in various cities. If we want to achieve control of the waiting area, we have to bear the costs of OCR misrecognition, semantic analysis of text recognition, learning and recognition, and the price of misrecognition. In response to this situation, we try to solve the problem of waiting instructions such as Figure 7 As shown. In the figure, the left and right lanes of the intersection are green, the upper and lower lanes are red, and the vehicle enters the "waiting area" indicator. The edge of the "waiting area" corresponds to the boundary of the grid where traffic is prohibited. The corresponding position of the traffic control grid map is set to prohibit traffic. The "waiting area" can be used, right turns and U-turns are prohibited, and EVs are allowed to enter. When the green light passes, the grid corresponding to the edge of the "waiting area" is converted to the Idle state, allowing EVs to pass.

[0160] On the RM grid diagram, the grids used by participants are marked as Busy; the grids used by participants are maintained as Blocked; after a certain period of time, the Blocked grids are converted to Idle. Figure 3 , Figure 6It can be seen that the Busy and Blocked grids related to the participants are connected to form a gradient block, which is presented as an arrow-like figure. Through graphic algorithms such as Kalman, the speed and direction of the participants can be extracted. With the analysis of multiple continuous RM grid graphs, such as the use of genetic algorithms, more accurate related parameters can be obtained. The distances between participants and obstacles, between participants, and between participants and boundaries can also be measured through related image algorithms, and as the resolution of the grid increases, its accuracy will also increase. Blocked grids can be used to maintain a safe distance between participants. Through related algorithms, the movement trajectory of the participants can also be analyzed. The movement trend of the participants can be predicted. Moreover, the value of the grid is highly correlated with the probability of the participant in the grid, which can be directly used as the basis for SLAM, path planning, decision-making, control and other algorithms of the EV's OBU.

[0161] RMU centrally maintains and sends out military drumbeat-like signals to coordinate the pace of each EV's start, stop, acceleration and deceleration. The "drumbeat" is a signal that occurs according to a certain rule. The pace of each EV's action is related to the "drumbeat" frequency. The higher the frequency, the faster the EV's movement and the faster the RM grid map is updated. The "drumbeat" is directly transmitted to the EV through V2I to unify the pace of each EV's action. For example, at an intersection, when the traffic instructions are about to switch, they will gradually slow down and simultaneously reduce the speed of each EV to be stopped; after the switch, they will gradually speed up and simultaneously increase the speed of the passing EV.

[0162] All data in the RM grid map are finite sets. Especially after the "drum beat" is used to unify the EV movement pace, the data in the RM grid map is similar and the values ​​are very concentrated. When the graphic compression algorithm is used to compress and transmit the plane, a very high compression ratio will be obtained, which will improve the transmission efficiency, save channel resources, reduce the interaction time, and improve the real-time performance of the system. The compression algorithms used include but are not limited to mature algorithms such as lzw, lz77, LZ78, H.264, H.265, jpeg, mpeg4, etc., and also include new algorithms customized for this map with more efficiency and higher compression ratio.

[0163] When EV moves on the road, it will occupy and release the unit grid road surface. Its behavior will be detected by the RMU, reflected on the RMU grid map, and fed back to the EV's OBU to form an information closed loop, thus realizing unmanned automatic driving. The whole process is as follows: Fig.10As shown. It is worth noting that the EV's communication equipment communicates with two RMUs, and traditional single-mode communication equipment can only communicate with one RMU at a time. For system safety, customization is recommended. The RMUs are arranged in a chain, and adjacent RMUs are covered by different wireless frequencies. The on-board communication unit can communicate with the corresponding RMU using different frequencies, which is technically feasible. For example, RMUA and RMUC use 4G networks, and RMU B uses 5G networks. When the EV is running in the first area, 4G communication equipment is used to communicate with RMU A, and 5G communication equipment is used to communicate with RMU B. When the EV is running in the second area, 4G communication equipment is used to communicate with RMUC, and 5G communication equipment is used to communicate with RMU B.

[0164] At the same time, the present invention also discloses an automatic driving implementation device, which includes a processor and a memory; the memory is used to store instructions, and when the instructions are executed by the processor, the device causes the device to implement any of the above-mentioned automatic driving implementation methods.

[0165] At the same time, the present invention also discloses a computer-readable storage medium, which stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer runs any of the above-mentioned methods for implementing autonomous driving.

[0166] In summary, the technical solution of this application provides a new solution for autonomous driving by integrating the operating system scheduling algorithm, managing the road surface, and reducing the probability of traffic accidents. At the same time, the operating system RM grid map is compatible with the autonomous driving planning and decision map. After simple processing, it can be used by the vehicle's OBU, reducing the complexity of the vehicle algorithm and improving the system's operating efficiency. By introducing synchronous drum beats (tick), attempts are made to solve the problem of inconsistent control steps of each EV, improve traffic efficiency, and reduce the probability of traffic accidents. Road monitoring is used to share holographic fresh environmental data for each vehicle, reduce system complexity, improve reliability, and improve system real-time performance; it provides the possibility of saving detection equipment for intelligent vehicles and reducing vehicle energy consumption and production costs. And the overlapping backup of road monitoring service areas improves system reliability. Engineering drawings can be used to reduce dependence on high-precision maps and reduce the dependence of unmanned vehicles on national geographical features. And the scheduling accurate to each grid unit is conducive to the refined management of the road surface and improves the compatibility of the autonomous driving system.

[0167] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for implementing automatic driving, characterized in that: include: Acquiring perception information, wherein the perception information includes environmental information, map information, and traffic control information; Based on the fusion processing of the perception information, the road management status, and the drum data, the autonomous driving related data is obtained; Based on the autonomous driving related data, road condition prediction, path planning and driving decision-making are performed, driving control is executed, and autonomous driving is realized; The autonomous driving related data obtained based on the fusion processing of the perception information, the road management status, and the drum data includes: After the roads are gridded, a grid data map is obtained; Based on the grid data map, an image algorithm is used to obtain a grid area related to the road surface as a region of interest (ROI); wherein each grid of the region of interest (ROI) corresponds to the right to use one grid unit of the road surface; Based on a preset calibration algorithm, the usage status of each grid is calibrated in combination with the perception information and the management status of the road; wherein the region of interest (ROI) is filled with a preset value; the road edge and unavailable grid cells are filled with a busy state or a non-state machine specific value; the available grid cells are filled with an idle state specific value to form a static grid map; Based on the real-time road conditions and traffic participant information in the service area, a preset algorithm is used to distinguish traffic participants and grids involved in collapse, excessive water, ice, dense fog, dense smoke, fire, and strong crosswinds, and calibrate them using specific values ​​consistent with the static grid map; among them, the road surface grids corresponding to traffic participants are calibrated using busy state specific values; the grids corresponding to road conditions such as collapse, excessive water, ice, dense fog, dense smoke, fire, and strong crosswinds are calibrated using busy state specific values ​​or non-state machine specific values ​​to form a dynamic grid map; Based on the perception information and the management status of the road, a traffic control grid map is generated, and traffic control signs are refined for each grid. The grid map of the service area is formed after being merged with the static grid map and the dynamic grid map, and is transmitted to the on-board unit (OBU) of the vehicle (EV) equipped with the communication system, so that the on-board unit (OBU) can realize road condition prediction, path planning, and automatic control decision-making; According to a preset drumbeat or a preset cycle, the usage status or time slice timing data of each grid in the grid map of the region of interest (ROI) is adjusted and maintained; Based on the movement of the vehicle (EV) equipped with the communication system, the occupation and release of the unit grid road surface in the grid map of the service area is updated, and the updated grid map of the service area is fed back to the on-board unit (OBU) of the vehicle (EV) equipped with the communication system to form an information closed loop. The on-board unit (OBU) of the vehicle (EV) equipped with the communication system realizes autonomous driving based on grid map prediction, path planning, automatic control, and driving decision of the updated region of interest (ROI).

2. The method for realizing automatic driving according to claim 1, characterized in that: The use status of calibrating each grid includes: The available road surface grid is initialized to an idle state. When a traffic participant enters a grid road surface, the corresponding grid is used and the state switches to a busy state. Other traffic participants no longer enter the road surface corresponding to the grid. When a traffic participant walks out of the grid road surface, the state switches to blocked, the grid is locked and waiting to be released, and the timer starts at the same time, and other traffic participants no longer walk into the road surface corresponding to the grid; After a certain period of time, the traffic participants have gone far away, and the road surface corresponding to the grid is safe and returns to the idle state, ready for use by other traffic participants.

3. The method for realizing automatic driving according to claim 1, characterized in that: The use status of calibrating each grid includes: The available road surface is calibrated by a state machine, and the unavailable road surface is calibrated by a busy state or a value different from that of the state machine.

4. The method for realizing automatic driving according to claim 1, characterized in that: In the static grid diagram and the dynamic grid diagram, the data of each grid is 8-bit data; the specific value of the state machine used contains three states and an unavailable specific value, which is represented by 2 bits.

5. A road-watching system for realizing autonomous driving, characterized in that: The method for implementing any one of claims 1 to 4 comprises: The perception unit is used to detect the environment, road surface, and traffic participants and obtain perception data; The computing unit is used to realize the fusion of perception data, the fusion of map data, communication data management, security data management, and computing services in edge computing; Through relevant algorithms, traffic participants and grids involved in collapse, excessive water, ice, and dense fog are distinguished; specific values ​​consistent with the static grid map are used for calibration, and the road surface grids corresponding to the participants are calibrated with busy state specific values; grids corresponding to road conditions such as collapse, excessive water, ice, dense fog, thick smoke, fire, and strong crosswinds are calibrated with busy state specific values ​​or non-state machine specific values ​​to form a dynamic grid map; The traffic control information detected by the perception unit, the traffic control information of the engineering drawing / map, and the traffic control information of the road monitoring system (RMU) covered by the autonomous driving platform management and configuration are verified with each other to obtain the traffic control information; The static grid map and the dynamic grid map are integrated into the traffic control grid map in the case of full data to form a grid map of the road monitoring unit (RMU) service area. The usage status or time slice timing data of each grid is adjusted and maintained according to the drum beat or fixed cycle. A communication unit for exchanging information with an intelligent transportation system (ITS) and a vehicle EV equipped with a communication system; The positioning unit is used to provide positioning services for the main body and EV equipment involved in autonomous driving; The clock unit is used to implement clock synchronization, high-precision time stamp, and drum beat in providing time series related services; Maintain and send out military drumbeat signals to coordinate the pace of starting, stopping, accelerating and decelerating of each vehicle (EV) equipped with the communication system; the drumbeat is a signal that occurs according to a certain rule, and the pace of each vehicle (EV) equipped with the communication system is related to the drumbeat frequency. The higher the frequency, the faster the movement of the vehicle (EV) equipped with the communication system, and the faster the road marking (RMU) grid map is updated; The storage unit is used to store engineering drawings, maps, operation logs, configuration parameters in non-volatile data, and to provide volatile data of operation memory and intermediate results for the computing unit; Security unit, used to provide security services such as encryption and decryption, communication authentication, backup voting and fault tolerance services, and equipment security monitoring services; The management unit is used to provide management services including configuration management, maintenance management operation statistics, status monitoring, fault warning and upgrade management services.

6. A road-watching system for realizing autonomous driving according to claim 5, characterized in that: Also includes: System interface, used to provide an interface for connecting to external systems.

7. A road-watching system for realizing automatic driving according to claim 5, characterized in that: Also includes: The power supply unit is used to provide power management and working power.

8. A road-watching system for realizing autonomous driving according to claim 5, characterized in that: The system adopts a chain-type overlapping arrangement.

9. A road-watching system for realizing autonomous driving according to claim 5, characterized in that: When the positioning unit fails, positioning is achieved using the trangle+ algorithm based on at least two of the systems.

Images